Multi-Attribute Generalization Method in Privacy Preserving Data Publishing

K-anonymization is a technique that prevents linking attacks by generalizing and suppressing portions of the released raw data so that no individual can be uniquely distinguished from a group of size of k. In this paper, we study single-attribute generalization for preserving privacy in publishing of sensitive data, and present multi-attribute generalization definition in process of generalization based Datafly algorithm. Besides we describe how to generate multi-attribute attribute generalization hierarchy. A key question is how to anonymize the microdata so that it can be multi-attribute attribute generalization hierarchy. We introduce the join, pune and create edge, as a way to reduce search space of k-anonymization, and propose a scalable and practical solution.